ARCHITECTURE OF AN INTELLIGENT SUBSYSTEM FOR DETECTING WARM-BLOODED PESTS IN MONITORING PRODUCTION PROCESSES IN AGRONOMY

<ns0:p>The paper considers the construction of an intelligent subsystem for detecting warm-blooded pests as a component of an automated system for monitoring production processes in agronomy. The relevance of the study is determined by the need for timely detection of biological threats in crop fields, which lead to significant economic losses and a reduction in the quality of the final product, a reduction in dependence on manual field inspection, and an improvement in the promptness of decision-making in precision agriculture. Based on the analysis of current approaches to computer vision, thermal imaging, drone monitoring, and IoT infrastructure, the requirements for a subsystem architecture are formulated. The subsystem should provide multimodal data acquisition, preprocessing, localization of potentially dangerous objects, threat assessment, and transmission of results to the decision-support loop. The purpose of the study is to substantiate the structure of such a subsystem, identify its main functional modules, and determine suitable classes of object detection models for field conditions. The study uses methods of system analysis, functional decomposition, comparative analysis of object detector architectures, and generalization of recent publications in precision agriculture. As a result, a multi-level architecture is proposed that combines the sensing layer, the analytical processing layer, the spatial verification layer, and the decision-support layer. The expediency of combining RGB and thermal data, as well as the use of YOLO-type one-stage detectors for real-time field detection, is substantiated. The practical value of the proposed approach lies in the possibility of its further integration into automated agricultural systems and the development of specialized early-warning services.Received 2026-04-01Accepted 2026-04-20</ns0:p>